JPH09205458A - Circuits and methods for intelligent response-based flow control in processing system networks - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method of the flow control of an intelligent response base in a processing system network. SOLUTION: A detection circuit 540 monitors first standby time characteristics for indicating at least a part of the utilization rate level of the network and monitors second standby time characteristics for indicating at least a part of an efficiency level relating to the transmission of a reception certification mark by a transmission circuit 515. A control circuit 545 makes a management circuit 510 able to manage re-transmission delay as the function of the utilization rate level of the network by adjusting the delay of the re- transmission of the data packets of the network 100 as the function of the first standby time characteristics and makes the management circuit 510 manage transmission delay as the function of the efficiency level relating to the transmission circuit 515 by adjusting the delay of the transmission of the reception certification mark as the function of the second standy time characteristics.

Description

Detailed Description of the Invention

[0001]

FIELD OF THE INVENTION The present invention relates to network connectivity, and more particularly to circuits and methods for intelligent response-based flow control in processing system networks.

[0002]

A processing system network, such as a computer network or communication network, couples two or more independent nodes capable of communicating with each other over a communication channel, path or link. It was done. The nodes are
It is also an independent processing system, such as a conventional computer or other processing system network.

Nodes communicate with each other to share resources such as databases and data files, software applications, hardware peripherals and communication links. Communication link is not directly linked 2
Shared to allow one or more nodes to communicate via one or more intermediate nodes. Resource sharing typically involves moving large amounts of data. The data is illustratively divided into packets, frames, groups, etc. (data packets). Each data packet contains data or information that is routed between two or more nodes.

Networks traditionally belong to one of two general categories: local area networks (LANs) and wide area networks (WANs).
A LAN is a group of communication nodes that are located relatively close to each other, such as in the same building or building complex. A WAN, on the other hand, is a collection of independent and separate network nodes that collaborate over a relatively long distance. Communication links between WAN nodes are routinely provided by third party carriers, such as long distance carriers. Router, bridge or other suitable network portal device
gateways such as devices), LANs and WANs
To connect together (eg LAN-LAN, L
AN-WAN, WAN-WAN) is used. Traditional approaches use gateways as junction points that route data packets received from a source network to one or more destination networks. The gateway typically includes control circuitry and memory. The memory often contains an address for routing to each LAN and WAN connected to the gateway, and also contains the address of one or more nodes of each LAN and WAN. If a particular node is itself a LAN, then one or more addresses of that node are often stored in memory as well.

Communication between the network and the nodes involves level manipulation. Level manipulation includes flow control of data traffic, sequencing, transmission error detection and correction, and the like. For example, consider three sequentially coupled networks (network A, network B, network C) in which one or more data packets are transmitted from a node in network A to a node in network C. It should be noted that the data packet passes from the network A to the network B and then to the network C like a baton in a relay race until it reaches a target node.
The target node processes the received packet to determine if it was received correctly. In response, the target node illustratively means an "ACK" signal, which means that the transmission was received without error,
Alternatively, it returns a "NAK" signal which means that the transmission is corrupted. AC on the transmitting node
Upon receiving the K signal, the node purges the previously transmitted data packet from its transmission queue. Alternatively, if the node receives a NAK signal, the transmitted node retransmits at least one or more data packets to the target node.

A timer is illustratively used to check for the occurrence of an event within a predetermined time period, such as the receipt of one or more data packets. The transmitting node, when transmitting a data packet, starts a transmission timer set to expire within a set period when the target node fails to respond in the form of an ACK or NAK signal. . If it expires, the transmission is retried and the transmission timer is reset. However, a common problem with the value of the transmission timer is that it fails to account for increases or decreases (ie, channel utilization) within the traffic of the network. As the number of tracks increases, time-outs tend to occur later. As another issue related to the destination node, after returning to the ACK, the node often needs to return one or more data packets to the transmitting node. In other words, after processing the received data packets, the destination node transmits to one or more obtained data transmitting nodes, which increases the overall network traffic.

A solution to this latter problem is to acknowledge the node of interest for a sufficient period of time to process the received data packet and to "piggyback" the ACK on the data packet to be returned. Delaying the return of a dynamic response. Illustratively, the destination node, upon receiving the data packet, has a response timer set to expire if the destination node fails to generate a return data packet within a set period of time. Let it start. When it expires, the target node transmits an ACK signal. A common problem associated with response timers is the inability to account for a series of data received at a destination node that does not generate return data. That is, the response timer tends to repeatedly time out, and the transmission of the ACK signal to the transmitting node is unnecessarily delayed. Unfortunately, these repeated timeouts slow down the transmission mode and increase the occurrence of transmission timeouts and unnecessary retransmissions if properly concatenated with heavy traffic on the network. Will end up.

Therefore, in the prior art, it has been necessary to effectively limit the occurrence of response-based timeouts, in particular, timeouts that result from increased network traffic. Also, in the prior art, there has been a need to more efficiently retransmit lost data packet transmissions that occur during intervals when network traffic decreases. Further, the prior art identifies a series of received data packets that will probably not return data to the transmitting node, and also modifies the response timer appropriately to respond more efficiently to the reception of the data packets. There was a need.

An object of the present invention is to dynamically and adaptively improve the efficiency of data packet transmission in order to solve the above problems.

[0010]

According to one feature of the invention, in a method for managing transmission of a signal by a transmission circuit over a network, an efficiency level associated with the transmission of the signal by the transmission circuit, Alternatively, monitoring latency characteristics indicative of at least a portion of any of the utilization levels of the network, and adjusting the transmission delay of the transmission circuit for the signal on the network as a function of latency characteristics. A method is provided that is characterized by managing transmission delay.

According to another characteristic of the invention, in a management circuit for managing the transmission of data packets and reception indicia by the transmission circuit over the network, at least a part of the utilization level of the network is indicated. A detection circuit operative to monitor a latency characteristic of No. 1 and a second latency characteristic indicative of at least a portion of an efficiency level associated with the transmission of a reception indicium by said transmission branch; Associated with the detection circuit and the transmission circuit, the management circuit adjusting the delay of the transmission circuit's retransmission of the data packet on the network as a function of a first latency characteristic. Allowing the delay to be managed as a function of the utilization level of the network, and of the receipt indicia on the network Adjusting the transmission delay of the sending circuit as a function of the second latency characteristic operates to allow the management circuit to manage the transmission delay as a function of an efficiency level associated with the transmission circuit. There is provided a management circuit having a control circuit.

[0012]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the accompanying drawings.

Referring to FIG. 1, a block diagram of the architectural structure of an exemplary TCP / IP network 100 is illustrated. TCP / IP is a set of protocols developed for linking dissimilar computers through many types of networks. TC
P / IP was used herein for exemplary purposes only. In practice, the principles of the present invention are based on Open Systems Interconnection (OSI), X. It is implemented according to any suitable response-based data packet transmission scheme, such as 25. "Including" is used herein to mean "including without limitation".

The exemplary network 100 includes a plurality of interconnected nodes 105. The first set of nodes 105 form a first sub-network 110a and the second set of nodes 104 form a second sub-network 110. Exemplary subnetwork 11
0 are coupled via an exemplary gateway 115 (ie, router, bridge or other suitable portal device). “Or” as used herein “and / or”
Means The gateway 115 is preferably a core gateway or a non-core gateway. The core gateway contains information about the structure of the network 100, and the non-core gateway contains limited or incomplete routing information, ie, one or more routing information to the core gateway. , Does not include any other routing information.

The core gateway typically stores the routing information in one or more routing tables. Each node 105 only needs to know the specific route to its local gateway 115.
Data packets transmitted between two distant nodes are therefore passed from one gateway to another until it reaches the destination node.

The routing table is preferably defined by the programmer, or dynamically refined (ie, the gateway queries neighboring gateways, etc. for routing information).
It preferably includes routing information for directly attached networks, as well as general knowledge for routing traffic to remote networks. If a gateway cannot resolve a particular routing address, it may ask other gateways for request seek assistance in this regard.
ng assistance).

Generally, when a data packet is routed between locations across multiple subnetwork environments, it is processed to determine the address of its intended location, or intermediate location. In response, the data packet is preferably encased or wrapped in the data envelope. Data envelopes, such as file transfer protocol frames, typically include a header with routing and transmission information. The TCP header provides, for example, end-to-end confirmation or response.

With reference to FIG. 2, a block diagram for routing one or more data packets between first and second locations within an exemplary TCP / IP network 100 is provided in this book. Shown according to the invention. The exemplary network 100 includes a plurality of nodes 105, each locally coupled to a suitable sub-network 110. First
Node 105a is connected to the Internet protocol (I
Acting as a host of P), it uses the IP addressing scheme to transmit the exemplary data packet 205 to the destination node 105n.

The first sub-network 110a processes the data packet 205 and also determines that the destination node 105n is not local. In response to this, IP
A header is added to the data packet. This IP header is the global I for the second sub-network 110n.
It contains the P address. The gateway 115 routes the data packet 205 based on the address of the second sub-network 110n. The second sub-network 110n determines whether the same address of the target node 105n is local or remote. If it is local, the data envelope is stripped and the data packet 205 is sent to the target node 10 as illustrated.
Routed to 5n.

If the destination node 105n is not local, the network 110n will optimally determine the address of the next network element, and the existing IP at that address.
Add a data packet. When an element receives a data envelope containing an IP data packet, it strips a header specifying its own address, and does the same as the previous network did, whether the destination address is local or not. IP in the header for other addresses for the next network element
Package the data. This process continues until the data packet is finally routed to the destination address.

In the illustrated embodiment, the first node 105a further functions to retransmit the data packet in response to a transmission timeout. The term "timeout" is used herein to include an appropriate comparison between a threshold and a timer or clock value that produces a particular result. The target node 105n operates according to TCP for inspection and preferably corrects the received data packet 205 if necessary or possible. Target node 1 if correct
05n returns a response signal to the first node 105a, otherwise returns a negative response appropriately.

The destination node 105n also suitably uses a response timer to measure the elapsed time for receipt of the data packet 205. “Measure” as used herein includes comparing, counting, equalizing, evaluating, gauged, inspected, weighted and the like. If the data packet 205 is processed and one or more data packets are to be returned to the first node 105a, the destination node piggybacks a response such as an ACK to at least one data packet. . The target node 105n preferably further operates in response to the response timeout to transmit only the response to the first node 105a.

The exemplary transmission timer value is non-limiting, but preferably depends on one or more of the following:

(A) Round-trip propagation delay of the signal (usually a small value, except for very long and very fast circuits); (b) Processing time at the receiver (including data packet latency). ); (C) Response signal or frame transmission time; or (d) Waiting and processing time at the transmitter when a response is received.

Here, due to the heavy load of the network traffic, the reception of the response signal by the first node 105b is delayed, which causes a timeout and the retransmission of the data packet 205. First node 1
05a, in the illustrated embodiment, operates to monitor the frequency of occurrence of transmission timeouts. If this is inversely compared with a threshold (compare unfavorably),
The first node 105a preferably changes the timeout function (eg, changes the timeout threshold, slows the timer or clock, etc.), temporarily disables the timeout function, and temporarily requests a response. It works like disabling. The term "impossible" is used herein to separate and disassociate (disassociate).
associate), disconnect (disconnect),
Disengage, pair (im
pair), incapacitate (incapac)
Itate), separation (separate) and the like. An important feature of the illustrated embodiment is that the efficiency of data packet transfer is improved.

The target node 105n operates in the illustrated embodiment to monitor the frequency of occurrence of response timeouts. When comparing this inversely with a threshold, the first node 105a preferably modifies the timeout function (eg, modifies the timeout threshold, slows the timer or clock, etc.) and It temporarily disables the time-out function and temporarily disables the response request. Another important feature of the illustrated embodiment is improved response transfer efficiency.

With reference to FIG. 3, an exemplary processing system 1
05 is illustrated. The processing system 105 can function as a node within either the exemplary processing system of FIG. 1 or FIG. The processing system 105 includes a monitor 305, a housing 310 and a keyboard 315.

The housing 310 includes a hard disk drive 320 and a floppy disk drive 325. Hard disk drive 320 is suitable for providing high speed storage and retrieval. Floppy disk drive 325 operates to receive and read or write to an external disk. Floppy disk drive 325 includes tape and compact discs, telephone systems and devices (including telephones, videophones, facsimile machines, etc.), message paging, network communication ports, and other conventional means for transmitting data or instructions. Replaced or combined with structures.

The housing 310 is partially cut away to illustrate the battery 330, the clock 335, the processor 340 and the disconnected local memory 345, all properly housed inside. Although the processing system 105 is illustrated as having a single processor, a single hard disk drive and a single local memory, the processing system 105 may be provided with multiple or combined processors or storage devices. The processing system 105 actually includes videophones, televisions, pagers, smart calculators, and handhelds, laptops / notebooks, minis, mainframes and supercomputers, and the like, as well as processing system networks combining these, Any suitable processing system operating in accordance with the principles of the present invention may be substituted or combined.

The architecture of a conventional processing system is
MacM by William Stallings
Illan Publishing Co. (3rd
ed. 1993) Computer Organi
Zation and Architecture. A conventional processing system network design is described in Darren L. et al. McGraw Hill, Inc., by Spohn. D of (1993)
It is described in detail by ata Network Design. In addition, conventional data communication is based on R. D. G
itlin, J .; F. Hayes and S.H. B. Wei
Plenum Press by Nstein (19
92) Data Communications P
rinciples, and James Harry
Irwin Professionala by Green
l Publishing (2nd ed. 199
2) The Irwin Handbook of
It is described in detail by Telecommuniations. These documents are incorporated herein by reference.

Referring to FIG. 4, a block diagram of an exemplary microprocessing system 400 suitable for implementation in the processing system of FIG. 3 is illustrated. Microprocessing system 400 includes a processor 340 connected via data bus 405 to an attached local memory 345. Memory 345
Operates to store data or instructions for the processor 340 to retrieve and execute.

The processor 340 is a control unit 410,
It includes an arithmetic logic unit (ALU) 415, as well as an internal memory 420 (eg, stackable cache, registers, etc.). Control unit 410 operates properly to fetch one of the instructions from memory 345. ALU 415 operates to perform the multiple operations necessary to execute these instructions, such as append and boolean AND. Internal memory 420 operates properly to provide local, fast storage used to store temporary results and control information.

Referring to FIG. 5, an exemplary network interface 500 (eg, modem card, FDD
High level block diagrams of I cards, Ethernet cards, etc.) are illustrated in accordance with the principles of the present invention. The interface 500 includes an exemplary memory 505, a management circuit 510, a transmission circuit 515, an exemplary expansion bus connector 520, and an exemplary connector 5 for twisted wires.
25, an exemplary coaxial cable connector 530, as well as an exemplary transmitter connector 535. The management circuit 510 includes a detector circuit 540 and a control circuit 545.

The interface 500 is properly connected to the exemplary processing system 105 of FIG. The management circuit 510 includes a plurality of data packets and one of the receiving indicia such as ACK and NAK by a transmission circuit 515 on a network such as the network 100 of FIGS.
Operates to manage one or more transmissions. The management circuit 510 includes a detector circuit 540 and a control circuit 54.
5 is comprised.

The detector circuit 540 operates to monitor the first latency of the network. The first waiting time characteristic indicates and represents at least a part of the utilization rate of the network. Network utilization levels are generally important for data packet transmission, and more specifically for data packet retransmission. As mentioned above, the threshold associated with the transmission timer is important for a valid retransmission.
The detector circuit 540 also operates to monitor the second latency characteristic. The second latency characteristic is of the efficiency level associated with the transmission of the receipt indicia by the transmission circuit 515,
It shows at least a part. The efficiency level associated with the transmission of the received indicia is generally important for data packet responses, and more specifically for positive responses and ACK piggybacks. As mentioned above, the threshold associated with the response timer is important for efficient data packet response.

The control circuit 545 is associated with both the detector circuit 540 and the transmission circuit 515. The control circuit 545 operates to properly adjust the retransmission delay on the network. Retransmission delay is the time interval between one or more transmissions and subsequent retransmissions. The adjustment is performed as a function of the first delay time characteristic. This allows the management circuit 510 to manage the retransmission delay as a function of network utilization level.

The control circuit 545 further operates to properly adjust the transmission delay associated with the retransmission of the indicia on the network. This transmission delay is the time interval between receipt of a data packet and the resulting receipt indicia without piggybacking the returned data packet.
The adjustment is performed as a function of the second delay time characteristic. As a result, the management circuit 510 reduces the transmission delay by the transmission circuit 515.
Can be managed as a function of the efficiency level associated with.

In carrying out either of the above two adjustments, the control circuit 545 controls the applied mathematical theory, including statistics, probability models, chaos theory, standard deviation, probability theory, permutations and combinations, frequencies, etc. The principle can be used properly. Furthermore, in addition to using timers, clocks, etc. in deriving the first measurable characteristic,
Physically detectable characteristics or features that impact communication throughput can be used appropriately (eg, communication channel utilization indicia).

Although network interface 500 is used to exemplify one circuit embodiment of the present invention, other circuit configurations may be properly implemented in other processing systems, nodes, gateways, etc. . In particular, in another embodiment, the circuit described above,
The microprocessing system 400 of FIG.
Proper replacement or combination of (programmable array logic), PLA (programmable logic array), DPS (digital signal processor), FPGA (field programmable gate array), ASIC (dedicated integrated circuit) VLSI (very large scale integrated circuit), etc. You can

Referring to FIG. 6, a flow diagram for performing intelligent response based flow control in accordance with the principles of the present invention is shown. For illustration purposes,
The description of FIG. 6 will be given with reference to FIG. Here, the management circuit 510 and the transmission circuit 515 may be located in the node.

A determination is made whether a data packet has been received from a transmitting node, gateway, etc. (decision step 602). If a data packet is received (decision step 602, YES branch), the transmission circuit 515 is disabled, the response timer is reset, and the response timer starts generating a transmission delay (process step 6).
04). Transmission delay represents the elapsed time interval for a received data packet. A determination is made whether the transmission delay was inversely compared to the transmission threshold (decision step 6).
06).

If the transmission delays are compared in the opposite way (YES branch of decision step 606), a reception indicia, such as a signal, is returned to the transmission node, gateway, etc. The detector circuit 540 derives the latency characteristic as a function of the comparison (process step 610) and operates to compare it with a latency threshold, eg, the maximum frequency of occurrence (process step 610). 612). If the characteristics of the waiting time are compared in reverse (YES in decision step 614)
Branch), the control circuit 545 operates to adjust or change the response timeout function (eg, change the timeout threshold, slow the timer or clock), temporarily disable it, and request a response. Is temporarily disabled (process step 616), which improves the overall efficiency of the response move.

In other words, the management circuit 510 operates to monitor the characteristics of the latency and to adjust the transmission delay of the transmission of the reception indicia on the network. This is done as a function of latency characteristics.

If no data packet is received (NO branch of decision step 602), a determination is made as to whether the data packet was transmitted (decision step 618). If the data packet is to be transmitted (YES branch of decision step 618), the transmission circuit 515 is enabled, the retransmission timer is reset, and the retransmission timer starts the occurrence of a retransmission delay (process step 6).
20). Retransmission delay represents the elapsed time interval for the transmission of a data packet without receipt of a reception indicia, such as a response signal. A determination is made whether the retransmission delay inversely compares the retransmission threshold (decision step 62).
2).

If the retransmission delays are compared inversely (YES branch of decision step 622), the data packet is retransmitted on the network (input / output step 62).
4). Detector circuit 540 operates to derive latency characteristics as a function of comparison (process step 61).
0) and also this, for example, the maximum frequency of occurrence,
Compare with latency threshold (process step 61
2). If the latency characteristics are compared in reverse (YES branch of decision step 614), the control circuit 545 adjusts or modifies the retransmission timeout function (eg, modifies the timeout threshold, slows the timer or clock). And temporarily disable it, process the retransmission request, and so on (process step 626), which improves the overall efficiency of the retransmission of the data packet. To be done.

The principles of the present invention illustrated above can be properly implemented as software. Exemplary software embodiments include conventional storage media having multiple instructions stored therein. These instructions can be read and executed by a suitable processor. These instructions, when executed, operate to control the processor to route data packets through the processing system network in accordance with the present invention. Preferred storage media include, but are not limited to
Magnetic, optical, and semiconductor, and suitable combinations thereof. A preferred software embodiment for managing the transmission of receipt indicia by the transmission circuit is UNIX.
NCR's STARGROUP® LAN Manager-OSI Trans for
It forms part of the Portion Version, which is the NCR Corpo in Dayton, Ohio.
available from ration.

[Brief description of drawings]

FIG. 1 is a block diagram of an architectural structure of an exemplary Transfer Control Protocol / Internet Protocol (TCP / IP) network.

FIG. 2 is a block diagram for routing one or more data packets between first and second locations in a network of another exemplary processing system.

FIG. 3 is an illustration of an example processing system that can function as a node in a processing system network.

4 is a block diagram of an exemplary microprocessing system suitable for incorporation within the processing system of FIG.

FIG. 5 is a high level block diagram of an exemplary network interface in accordance with the principles of the present invention.

FIG. 6 is a flow diagram for performing intelligent response-based flow control in accordance with the principles of the present invention.

[Explanation of symbols]

 115 gateway 205 data packet

 ─────────────────────────────────────────────────── —————————————————————————————————————— Inventor Jeffrey M. Donnelly, New Jersey, USA 07733 Holmdel, Armand Court 16

Claims (2)

[Claims] 1. A method for managing the transmission of a signal by a transmission circuit (515) on a network (100), wherein either the efficiency level associated with the transmission of the signal by the transmission circuit or the utilization level of the network. Of the latency characteristics exhibiting at least a portion of the delay time, and managing the transmission delay by adjusting the transmission delay of the transmission circuit for the signal on the network as a function of the latency characteristics. And how to. 2. A management circuit (50) for managing the transmission of a data packet (205) and a reception indicia by a transmission circuit (515) on a network (100), the network utilization level of Monitoring a first latency characteristic exhibiting at least a portion and a second latency characteristic exhibiting at least a portion of an efficiency level associated with the transmission of the indicia by the transmission branch. And a detection circuit (540) operating in accordance with the above, and adjusting the transmission circuit retransmission delay of the data packet on the network as a function of a first latency characteristic. The management circuit controls the retransmission delay as a function of the utilization level of the network. In addition, the management circuit adjusts the transmission delay of the transmission circuit of the reception indicia on the network as a function of the second latency characteristic so that the management circuit delays the transmission delay with respect to the transmission circuit. Management circuit characterized in that it has a control circuit (545) which can be managed as a function of the level.